Reaction cascades of central metabolic pathways are well characterized, yet still relatively little is known about their regulation and coordination. However, homeostasis of cellular metabolites and metabolic fluxes is essential for cellular growth, and needs to be adapted to the growth conditions of all living cells [3
] Recently, it has become clear that the metabolic enzyme PK is required for proper coordination of energy- and redox metabolism. In its glycolytic role, PK converts phosphoenolpyruvate (PEP) to pyruvate, yielding a molecule of ATP, and thus, this reaction is required for the net ATP production in glycolysis [33
]. Interest in PKM2 had increased markedly after it had been suggested that PKM2 expression in adults could be specific to cancer cells, and that a PKM1 to PKM2 isoform switch might be responsible for increased fermentative metabolism (Warburg effect) of cancer cells [34
]. Cancer specificity of PKM2 could however not be confirmed in recent studies, as PKM2 was found to dominate over PKM1 in most adult tissue, including kidney, lung, liver, thyroid, bladder and colon [20
], except however muscle [34
]. Despite not specific to cancer, several studies established an essential role of PKM2 in metabolic regulation and the oxidative stress response, processes of high importance for cancer development [17
]. First, the literature appears consistent that PKM2 is upregulated on transcript or protein level in several malignant and benign tumors. These studies included human melanoma [41
] childhood leukemia [42
], renal cell carcinoma, bladder carcinoma, hepatocellular carcinoma, colorectal carcinoma, lung carcinoma, thyroid follicular adenoma [20
], and mouse kidney tumors [36
]. Several cancer tissues also expressed the PKM1 isoform, however at much lower levels than PKM2. In some but not all tumors, also the PKM1 isoform was found to be upregulated [20
Conversely, at the same time it has been reported that the PK activity is reduced by posttranslational modifications in cancer cells, and that this reduction is required for their proliferation [18
]. Reduced PK activity was attributed to PKM2 phosphorylation [19
], thiol oxidation [18
], and acetylation [43
]. This indicates that PKM2 is upregulated in cancer to increase the protein content, but not to increase the overall PK enzymatic activity. However, it has also to be noted that it is yet unclear to which extent the identified modifications might also be found on PKM2 in normal human tissue, and thus, and as to which extend they are cancer cell specific.
Here, we followed the proteome wide consequences of the Pyk1p downregulation by creating shotgun proteome of yeast expressing Pyk1p at high and low levels using a latest-generation QqTOF proteomics platform, a TripleTOF™ 5600 (AB SCIEX) mass spectrometer coupled to a high-pressure nanoLC system was used for a label free proteome profiling. Using the paragon research engine, we identified >900 proteins per single run. A set of 393 proteins met the stringent criteria for being quantified, i.e. that a minimum of three specific peptides were identities in all 19 shotgun runs. The proteome profiles clearly clustered dependent on the PK expression level (Fig ) This clear separation was caused by 7.8 percent (31) proteins that were significantly responding to changing PK levels.
These numbers illustrate both the advantages and disadvantages of current label free proteomic approaches. On the plus side, label-free profiling is much more flexible and convenient, as isotope labeled standards are costly, and these standards can only be created for a limited number of species. In this particular case, i.e. SILAC labeling was difficult, as the yeast genetic modifications necessary [44
] would impact yeast amino acid transport and metabolism, and thus potentially interfere with the biological function of PK. On the down side, at the current state of the art, the number of proteins quantifiable in a labeled experiments is higher, as the peptide pools can easily be pre-separated before the LC runs. However, also when using isotope labels, the number of replicate injections may be limited by the throughput of the nanoLC gradients (i.e. the 19 injections using at a 3hr gradient required one week of instrument time). These current bottlenecks for label-free approaches might however improve in the near future, as recent developments in data-independent acquisition (i.e. SWATH [45
]) will facilitate significant faster runtime while increasing in the number of quantifiable peptides and proteins.
To verify the proteome profiles, we applied the targeted MS technique selective reaction monitoring (SRM) and analyzed the expression of nine randomly chosen PYK targets. Facilitated by the shorter runtimes and the high reproducibility of the SRM experiments, we included three additional strains that have different PK activities, the wild type parent BY4741, as well as two yeast strains expressing the PYK2
both at high and low level [17
]. For seven of the nine tested proteins, this extended study did confirm Pyk1p, and also Pyk2p dependent regulation. Furthermore, the SRM experiments allowed grouping of these PK targets according to their co-expression pattern. Remarkably, co-expressed proteins formed functional clusters: proteins up-regulated were glycolytic enzymes, proteins downregulated were involved in translation. PK dependence of glycolytic enzyme expression is an interesting parallel to the humans in respect to cancer development: hydroxylated PKM2
leads to an upregulation of glycolytic enzymes through interacting with hypoxia inducible factor -1α (HIF1α) [38
]. Thus, in both species regulation of glycolytic enzyme abundances can be attributed to PK, and hence, at least also parts of non-catabolic functions of PK appear to be conserved between yeast and humans. Interestingly, we also observed overlapping targets upon an activation of the PPP during the oxidative stress response [16
]. As low PYK activity increases the concentration of PPP metabolites [17
], it is conceivable that the PPP is involved in gene expression regulation mediated by PK. In this context is has to be noted that switches of glycolysis to the PPP can howeveronly partly attributed to transcriptional regulation, i.e. metabolic shifts in glycolysis and the PPP can be regulated on the metabolic level only [46
], and may occur much faster than transcriptional regulation [47
The vast majority of identified PK targets were involved in metabolic processes related to glycolysis, the synthesis of nucleic acids, amino acids and proteins. Although it is assumed that PK mediates the regulation of glycolysis [39
], similar mechanisms might also exist for amino acid metabolism. To obtain a metabolic picture of the global PK function in this process we thus set out to use ion exchange chromatography and ninhydrin detection to profile the levels of free amino acids in the PK model strains. Comparing yeast with different PK expression level and activity, we found a clear reconfiguration of the overall amino acid profile. In cells with low PK activity, seven of 17 amino acids were significantly less concentrated, and two were higher concentrated.
Amino acid levels are important indicators for the metabolic homeostasis of cells – and closely interconnected with protein metabolism. In yeast, mammalian cells, and flies, proteasome inhibition is lethal due by causing amino acid shortage [48
], implying that maintaining sufficient amino acid supply is crucial for proliferating cells and tumors. One of the amino acids we found at decreased concentration was serine. Remarkably, it has recently been described for human cells that PKM2 stimulates de novo serine biosynthesis [49
], and that in turn serine can act as allosteric activator of PKM2 [50
]. This implies a feedback circle involving PYK that controls serine production in mammalian cells, providing evidence that the PK's role in regulation of serine metabolism is conserved between yeast and mammals.
Only two amino acids increased in concentration when there was low PYK activity, glutamine and glutamate. Turnover of these two amino acids may be of crucial importance for cancer metabolism, as cells can obtain a substantial amount of their energy through the process of glutaminolysis [3
]. A change in pyruvate kinase expression seem thus shift energy metabolism towards glutaminolysis, providing sufficient energy supply for cancer cells.
To test for conservation in humans, we searched for human orthologues of the proteins regulated by yeast PK. Remarkably, two direct orthologues were known to be associated with cancer progression. The first protein, asparagine synthase, is the orthologue of yeast protein ASN1, ASNS (Fig ). ASNS is as potential target for the treatment of castration-resistant stage of prostate cancer [52
]. Moreover, the expression level of this enzyme is an indicator for the efficiency of the enzyme drug L-Asparaginase (L-ASP) in ovarian cancer [53
]. The second protein, serine hydroxymethylstransferase (SHMT1), is the human orthologue of the yeast protein SHM1
. SHMT1 catalyzes the formation of glycine and methylene tetrahydrofolate which are important precursors for nucleotide biosynthesis, and is discussed as a therapeutic target since the late 1980s [54
Taken together, these observations implicated that the function of PK in regulating amino acid metabolism is conserved between yeast and human. Therefore, we continued our investigations by quantifying further potential PK targets in a human tumor. For two reasons, we selected thyroid follicular adenoma for these investigations: First, these benign cancers are characterized by a clear upregulation of the PKM2
]. Indeed, normal and adenoma biopsies are clearly distinguished by the PKM2
], graphically illustrated in Fig ). Also PKM1
was expressed in these tumor tissues, but at around a 50 fold lower level. In difference to PKM2, PKM1
concentrations did not distinguish healthy thyroid from follicular adenoma tissue. Second, as adenomas are slow growing tumors, one can exclude unspecific effects which may result from altered metabolism of the rapidly proliferating cells. Finally, the activity of the respiratory chain complexes I – IV, determined in both the healthy as well as the adenoma biopsies, indicated significantly lowered activity of mitochondrial metabolism (Fig ). Thus, being characterized by increased PKM2
levels and a decrease in respiratory chain efficiency, thyroid follicular adenoma represented tissue exhibiting classic features of the ‘Warburg effect’
SRM assays were tested for all five human proteins, but only SHMT1 and the aminopeptidase AAP-S were detected at sufficient intensity for reliable relative quantification by our LC-MS/MS workflow. Remarkably, AAP-S and SHMT1 were both found clearly upregulated in these adenoma samples. At the measure of the relative expression level, AAP-S levels clearly distinguished adenoma and that of control tissue (p < 0.01). Plotted against the absolute concentration of PKM2, both targets created clusters that distinguished the adenomas from the control tissue. Thus, expression of two direct protein orthologues to the yeast PK targets correlate with PKM2 expression in thyroid follicular adenoma.